The way I produce the models that I 3D print allows me to work down to an accuracy of 0.001mm. Of course it's highly unlikely any printing process will be so accurate but so far when I've printed models in multiple pieces they have always fitted together perfectly. I was surprised therefore when I tried to put some of the second Clayton prototype together to find that I didn't have a perfect fit. The problem was that once I'd fitted the bearings to the FUD print the stainless steel keeper plate wouldn't fit without distorting the body. You can see this in the left hand photo.
What confused me was that the stainless steel part measured out at exactly 9.10mm on my digital callipers which matched the 3D model perfectly. As I mentioned before the stainless steel is seriously tough and it took ages, even with a diamond file, to reduce the width sufficiently to allow things to fit better (the right hand photo). Altering the 3D model to narrow the steel part slightly for future models is easy but I wanted to work out not only how much to narrow it by, but why I needed to.
When I develop a new model I produce virtual copies of all the other parts (wheels, bearings, motor, etc.) that I'll use to help me visualise how everything will fit together and to ensure adequate clearances etc. On this occasion though it appears I managed to wrongly measure the flange on the wheel bearings. My model of the bearings has the flange at 0.1mm, turns out having remeasured them that they are actually 0.2mm which means that the keeper plate is 0.2mm two wide. I know 0.2mm doesn't sound like much but I suppose that just goes to show how accurate the 3D printing process is.